Guile 3.0, again

[unspecd]

Testing the changes

• I tested the changes in this PR: YES

New package

• This new package conforms to the quality requirements: YES

Local build testing

• I built this PR locally for my native architecture, x86_64-musl
• I built this PR locally for these architectures (if supported. mark crossbuilds):
  • ppc64le-musl

[q66]

do we actually need this for anything

[q66]

also i'm opposed to enabling guile by default in any system packages, it's of little to use to anybody and maintaining the guile toolchain is enough of a pain in the ass i don't want it anywhere near base system (keep it as non-default build options or whatever, but no default guile dependencies please)

[ericonr]

do we actually need this for anything

Less so now that we have guile 2.2. But there are probably 3.0 things out there.

[unspecd]

Okay, it's a little better now, but not perfect: gnutls doesn't cross-compile. I'm working on it.

[sirikid]

do we actually need this for anything

GNU Guix

[unspecd]

GNU Guix

#22359 (comment)

[sirikid]

GNU Guix

#22359 (comment)

Long way to go, but we need Guile before Guix. Also, having Guix helps a lot in development.

[fosslinux]

Guix isn't in the repositories, I believe @q66 was talking about anything in the repositories that this is blocked on before an update or smth

w/r/t musl and guix, you'd need to source build packages as needed as was done in the link provided, as Guix has no musl support and hence no binaries.

[unspecd]

gdb works fine.

$ gdb -ex='handle SIGSEGV noprint nostop' -ex='b scm_display' -ex=r -ex=bt --args guile3 -c '(display "hello\n")'

Without guile:

#0  scm_display (obj=0x7ffff77a1920, port=0x904) at print.c:1110
#1  0x00007ffff7ebb4b9 in vm_regular_engine (thread=0x7ffff7910d80) at vm-engine.c:972
#2  0x00007ffff7ebdd39 in scm_call_n (proc=<optimized out>, argv=argv@entry=0x7fffffffd668, nargs=nargs@entry=1) at vm.c:1608
#3  0x00007ffff7e37027 in scm_primitive_eval (exp=<optimized out>, exp@entry=0x7ffff73e9560) at eval.c:671
#4  0x00007ffff7e37086 in scm_eval (exp=0x7ffff73e9560, module_or_state=0x7ffff73e6f00) at eval.c:705
#5  0x00007ffff7ebb4b9 in vm_regular_engine (thread=0x7ffff7910d80) at vm-engine.c:972
#6  0x00007ffff7ebdd39 in scm_call_n (proc=<optimized out>, argv=argv@entry=0x7fffffffd848, nargs=nargs@entry=1) at vm.c:1608
#7  0x00007ffff7e37027 in scm_primitive_eval (exp=<optimized out>, exp@entry=0x7ffff73dd100) at eval.c:671
#8  0x00007ffff7e37086 in scm_eval (exp=0x7ffff73dd100, module_or_state=module_or_state@entry=0x7ffff73e6f00) at eval.c:705
#9  0x00007ffff7e920f6 in scm_shell (argc=3, argv=0x7fffffffde58) at script.c:357
#10 0x00007ffff7e4f55d in invoke_main_func (body_data=0x7fffffffdd90) at init.c:308
#11 0x00007ffff7e308ba in c_body (d=0x7fffffffdcd0) at continuations.c:430
#12 0x00007ffff7ebb4b9 in vm_regular_engine (thread=0x7ffff7910d80) at vm-engine.c:972
#13 0x00007ffff7ebdd39 in scm_call_n (proc=<optimized out>, argv=argv@entry=0x7fffffffda70, nargs=nargs@entry=2) at vm.c:1608
#14 0x00007ffff7e35cfa in scm_call_2 (proc=<optimized out>, arg1=<optimized out>, arg2=<optimized out>) at eval.c:503
#15 0x00007ffff7e376bc in scm_c_with_exception_handler (type=type@entry=0x404, handler=handler@entry=0x7ffff7eb2bc0 <catch_post_unwind_handler>, handler_data=handler_data@entry=0x7fffffffdbf0, 
    thunk=thunk@entry=0x7ffff7eb2d00 <catch_body>, thunk_data=thunk_data@entry=0x7fffffffdbf0) at exceptions.c:170
#16 0x00007ffff7eb2f0d in scm_c_catch (tag=tag@entry=0x404, body=body@entry=0x7ffff7e308b0 <c_body>, body_data=body_data@entry=0x7fffffffdcd0, handler=handler@entry=0x7ffff7e30b60 <c_handler>, 
    handler_data=handler_data@entry=0x7fffffffdcd0, pre_unwind_handler=pre_unwind_handler@entry=0x7ffff7e309b0 <pre_unwind_handler>, pre_unwind_handler_data=0x7ffff77be540) at throw.c:168
#17 0x00007ffff7e30e78 in scm_i_with_continuation_barrier (body=body@entry=0x7ffff7e308b0 <c_body>, body_data=body_data@entry=0x7fffffffdcd0, handler=handler@entry=0x7ffff7e30b60 <c_handler>, 
    handler_data=handler_data@entry=0x7fffffffdcd0, pre_unwind_handler=pre_unwind_handler@entry=0x7ffff7e309b0 <pre_unwind_handler>, pre_unwind_handler_data=0x7ffff77be540)
    at continuations.c:368

With guile:

#0  scm_display (obj="hello\n", port=#<undefined 904>) at print.c:1110
#1  0x00007ffff7ebb4b9 in vm_regular_engine (thread=0x7ffff7910d80) at vm-engine.c:972
#2  0x00007ffff7ebdd39 in scm_call_n (proc=<optimized out>, argv=argv@entry=0x7fffffffd668, nargs=nargs@entry=1) at vm.c:1608
#3  0x00007ffff7e37027 in scm_primitive_eval (exp=<optimized out>, exp@entry=(display "hello\n")) at eval.c:671
#4  0x00007ffff7e37086 in scm_eval (exp=(display "hello\n"), module_or_state="#<struct module>" = {...}) at eval.c:705
#5  0x00007ffff7ebb4b9 in vm_regular_engine (thread=0x7ffff7910d80) at vm-engine.c:972
#6  0x00007ffff7ebdd39 in scm_call_n (proc=<optimized out>, argv=argv@entry=0x7fffffffd848, nargs=nargs@entry=1) at vm.c:1608
#7  0x00007ffff7e37027 in scm_primitive_eval (exp=<optimized out>, exp@entry=((@ (ice-9 control) %) (begin ((@@ (ice-9 command-line) eval-string/lang) "(display \"hello\\n\")") (quit))))
    at eval.c:671
#8  0x00007ffff7e37086 in scm_eval (exp=((@ (ice-9 control) %) (begin ((@@ (ice-9 command-line) eval-string/lang) "(display \"hello\\n\")") (quit))), 
    module_or_state=module_or_state@entry="#<struct module>" = {...}) at eval.c:705
#9  0x00007ffff7e920f6 in scm_shell (argc=3, argv=0x7fffffffde58) at script.c:357
#10 0x00007ffff7e4f55d in invoke_main_func (body_data=0x7fffffffdd90) at init.c:308
#11 0x00007ffff7e308ba in c_body (d=0x7fffffffdcd0) at continuations.c:430
#12 0x00007ffff7ebb4b9 in vm_regular_engine (thread=0x7ffff7910d80) at vm-engine.c:972
#13 0x00007ffff7ebdd39 in scm_call_n (proc=<optimized out>, argv=argv@entry=0x7fffffffda70, nargs=nargs@entry=2) at vm.c:1608
#14 0x00007ffff7e35cfa in scm_call_2 (proc=<optimized out>, arg1=<optimized out>, arg2=<optimized out>) at eval.c:503
#15 0x00007ffff7e376bc in scm_c_with_exception_handler (type=type@entry=#t, handler=handler@entry=0x7ffff7eb2bc0 <catch_post_unwind_handler>, handler_data=handler_data@entry=0x7fffffffdbf0, 
    thunk=thunk@entry=0x7ffff7eb2d00 <catch_body>, thunk_data=thunk_data@entry=0x7fffffffdbf0) at exceptions.c:170
#16 0x00007ffff7eb2f0d in scm_c_catch (tag=tag@entry=#t, body=body@entry=0x7ffff7e308b0 <c_body>, body_data=body_data@entry=0x7fffffffdcd0, handler=handler@entry=0x7ffff7e30b60 <c_handler>, 
    handler_data=handler_data@entry=0x7fffffffdcd0, pre_unwind_handler=pre_unwind_handler@entry=0x7ffff7e309b0 <pre_unwind_handler>, pre_unwind_handler_data=0x7ffff77be540) at throw.c:168
#17 0x00007ffff7e30e78 in scm_i_with_continuation_barrier (body=body@entry=0x7ffff7e308b0 <c_body>, body_data=body_data@entry=0x7fffffffdcd0, handler=handler@entry=0x7ffff7e30b60 <c_handler>, 
    handler_data=handler_data@entry=0x7fffffffdcd0, pre_unwind_handler=pre_unwind_handler@entry=0x7ffff7e309b0 <pre_unwind_handler>, pre_unwind_handler_data=0x7ffff77be540)
    at continuations.c:368

[q66]

i don't really care it "works fine", I just don't want a guile dependency anywhere near base-* packages

[unspecd]

I just don't want a guile dependency

Ok, what about python?

[q66]

python builds and works just fine on every known (and most unknown) target, guile is fragile, takes ages to bootstrap itself, and every version so far had an assortment of often arcane issues attached

[q66]

feel free to leave it as an option for those 2 people who are ever going to use it, but I don't want it burdening any of the base- packages, so no guile default options for things like gdb and make

[sirikid]

@unspecd can I help you somehow with this?

[unspecd]

can I help you somehow with this?

You could try to fix this :)

[sirikid]

can I help you somehow with this?

You could try to fix this :)

don't expect much, but I will try

[unspecd]

The bug is in the function scm_floor_divide in libguile/numbers.c:

void
scm_floor_divide (SCM x, SCM y, SCM *qp, SCM *rp)

On x86_64:

$ guile3
scheme@(guile-user)> (floor/ 130.0 10/7)
$1 = 91.0
$2 = 0.0

On i686:

$ guile3
scheme@(guile-user)> (floor/ 130.0 10/7)
$1 = 90.0
$2 = 1.4285714285714257

On i686, exact numbers:

$ guile3
scheme@(guile-user)> (floor/ 130 10/7)
$1 = 91
$2 = 0

-- good.

[unspecd]

Updated to 3.0.7.

[unspecd]

Updated guile-wisp.

[github-actions]

Pull Requests become stale 90 days after last activity and are closed 14 days after that. If this pull request is still relevant bump it or assign it.

[leahneukirchen]

State of guile support, September 2022: https://bbs.archlinux.org/viewtopic.php?id=279907

[daviwil]

How far is this from being done? Are the maintainers interested in this work being completed? I'd love to have Guix package manager officially packaged on Void.

[leahneukirchen]

I'd like to have Guile 3 too; packaging it in itself isn't the problem, but making sure all dependencies work.

Apparently Lilypond still needs Guile 2.2, but we could at least drop 1.8 then.